C Parallel merge sort working sometimes - c

Question

I'm learning how to parallize the merge sort in C following this tutorial
http://elc.yonsei.ac.kr/courses/csi2110/PP-L05-ScalableAlgorithmicTechniques.pdf But it only works sometimes. I run the code in the terminal about 10 times, and sometimes I get segmentation fault, other times, I get random numbers in my array, or sometimes, it works.
I'm not sure where I'm going wrong so any help would indeed be most appreciated.
#include <stdio.h>
#include <stdlib.h>
#include <pthread.h>
#include <math.h>
pthread_t LThread;
pthread_t RThread;
void MergeSort(float A[], int p, int r);
void ParallelMergeSort(float A[], int p, int r, int depth, int max_depth);
void Merge(float A[], int p, int q, int r);
struct arg {
float* A;
int p;
int r;
int depth;
int max_depth;
};
void* PMSort(void* ptr){
struct arg* MyArg = (struct arg*) ptr;
ParallelMergeSort(MyArg->A, MyArg->p, MyArg->r, MyArg->depth, MyArg->max_depth);
return 0;
}
void ParallelMergeSort(float A[], int p, int r, int depth, int max_depth){
if (depth == max_depth){
MergeSort(A, p, r);
}
else {
/*
1) Spawn 2 threads for left and right sub array
2) Join the 2 threads
3) Perform the merge
*/
int q;
if (p < r){
q = (p+r) / 2;
struct arg* LeftArg = malloc(sizeof(struct arg));
struct arg* RightArg = malloc(sizeof(struct arg));
LeftArg->A = A;
LeftArg->p = p;
LeftArg->r = q;
LeftArg->depth = depth + 1;
LeftArg->max_depth = max_depth;
RightArg->A = A;
RightArg->p = q + 1;
RightArg->r = r;
RightArg->depth = depth + 1;
RightArg->max_depth = max_depth;
pthread_create(&LThread, NULL, PMSort, (void*)LeftArg);
pthread_create(&RThread, NULL, PMSort, (void*)RightArg);
pthread_join(LThread, NULL);
pthread_join(RThread, NULL);
Merge(A, p, q, r);
}
}
}
void Merge(float A[], int p, int q, int r){
int n1 = q -p + 1;
int n2 = r - q;
int i = 0;
int j = 0;
int L[r];
int R[r];
for (i = 0; i < n1; i ++){
L[i] = A[p + i];
}
for (j = 0; j < n2; j ++){
R[j] = A[q + j + 1];
}
L[n1] = INFINITY;
L[n2] = INFINITY;
i = 0;
j = 0;
for (int k = p; k <= r; k ++){
if (L[i] <= R[j]){
A[k] = L[i];
i ++;
}
else {
A[k] = R[j];
j ++;
}
}
}
void MergeSort(float A[], int p, int r){
int q;
if (p < r){
q = (p + r)/2;
MergeSort(A, p, q);
MergeSort(A, p+1, r);
Merge(A, p, q, r);
}
}
int main(void){
float array[] = {5,2,4,7,1,3,2,6};
ParallelMergeSort(array, 0, 7, 0, 3);
for (int i = 0; i <= 7; i ++){
printf("%f ", array[i]);
}
printf("\n");
return 0;
}

Answer 1

Don't ignore return values from function calls in C. Start out by adding perror calls if your pthread calls do not return 0 and see what happens. You will see that the pthread_join calls are failing because you have RThread and LThread declared as globals. So you keep re-assigning them new values as you spawn off threads. Move those pthread_t declarations so they are declared inside the ParallelMergeSort function instead.
That won't fix any sorting issues with your algorithm but at least you will get consistent results.